Verified80% confidenceFactExact time
The Adam optimizer was proposed in 2014, combining momentum and adaptive learning rates, and has become the most widely used optimizer today.
7
Sources
80%
Confidence
Long-term
Relevance
7/19/2026
First Seen
Sources
梯度下降与反向传播详解：神经网络学习机制从零拆解
Related Claims
VerifiedAdaptive learning rate optimizers such as Adam and RMSProp further improve convergence speed and stability over SGD.80% similarUnverifiedAdamW、LAMB等优化器变体专门针对大批量训练和大模型微调进行优化74% similarVerifiedUsing the Adam optimizer with a default learning rate of 1e-3 as a starting point is a robust entry strategy for deep learning.73% similarUnverifiedMUON优化器替代了沿用多年的AdamW优化器，通过引入矩阵正交化步骤使训练曲线更加平滑67% similarUnverifiedAdamW maintains two exponential moving average states, the first moment and the second moment, for each trainable parameter, usually stored in fp32, and in mixed-precision training the optimizer states often take up more GPU memory than the model weights and activations.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/556690API
curl https://kongchang.com/api/v1/knowledge/claims/556690MCP
get_claim(id=556690)